Object-Oriented Processing of Structural Design Knowledge Using Neural Networks

摘要

The detailed stage of structural design has benefited considerably from computer automation. The most successful applications of this automation have been for the numerically intensive tasks of structural analysis and optimisation using the traditional procedural programming approach. However, engineers use more than just procedural knowledge when performing design; e.g., heuristic knowledge, knowledge of relationships, knowledge based on experience, etc., and many of these different types of knowledge do not lend themselves to a procedural representation. Albeit, in order to advance the state of automation of the detailed design process these other types of knowledge must be represented in some computational form. This paper describes how an object-oriented system for detailed structural design addresses this need, with particular emphasis on the use of artificial neural networks for extracting and then processing knowledge that is implicit in the solutions to previous design problems.